The equality paradox of deliberative democracy: Evidence from a national Deliberative Poll

Transcription

1 April 4, 2006 The equality paradox of deliberative democracy: Evidence from a national Deliberative Poll Assistant professor Kasper M. Hansen, Ph.D. University of Copenhagen Department of Political Science Øster Farimagsgade 5, P.O. Box 2099, DK-1014 Copenhagen K, Denmark kmh@ifs.ku.dk Phone Paper presented at ECPR-workshop no. 23: The role of political discussion in modern democracies in a comparative perspective Nicosia, Cyprus Abstract The focus on deliberative democracy has increased considerably within the last decade. So far the literature has primary praised deliberation for its great contributions to almost any political process. E.g., it has been argued widely that greater emphasis on deliberation would enhance the citizens capabilities to form reasoned and reflective opinions based on the common good rather than self-interest. The focus on the normative potential of deliberation has to a large extent neglected deliberations many pitfalls and contradictions. One of the most troubling paradoxes in deliberative democracy is the equality paradox. The equality paradox of deliberation relates to the idea that deliberative democrats advocate the importance of securing that the deliberative process fulfills political equality as the participants in the deliberative process are able to express their views freely and openly without any procedural restrictions. At the same time, however it is argued that arguments referring to the common good and public interest should and will be emphasized in a deliberative process, which favors participants accustomed to this kind of reasoning, thus, compromising political equality. Evidence from the latest national Danish Deliberative Poll supports the present of equality paradoxes of deliberation, which call for a revision of deliberative democratic theory in order to incorporate and give proper answers to the paradox. The revision of deliberation democracy suggests explicit to acknowledgment the trade-off between political equality and deliberation as a process based on argument which appeal to general interest and secondly increase the systematic use of moderated deliberation in order to provide a stronger opportunity for minorities to express their views.

2 Introduction The term deliberation can be traced to Latin, where Libra translates into to scale, weight or balance. In this sense, deliberation refers to de-weight, de-scale or de-balance, but is often simply translated into careful consideration. Explicitly deliberation can be defined as an unconstrained exchange of arguments that involves practical reasoning and potentially leads to a transformation of preferences (Cooke 2000:948; Hansen 2004:98). Within the last decades, deliberation has revitalized the theoretical debate within political and democratic theory, in applied empirical research as well as within the many new initiatives to involve citizens in public decision-making. In political theory Jürgen Habermas (1984; 1996b), Rawls (1971; 1996) and more recently Bohman (1996), Gutmann & Thompson (1996) and Dryzek (2005; 1990; 2000) have developed more or less embracing theories with strong deliberative elements. In the applied empirical research, James S. Fishkin and Robert Luskin s (1988; 1991; 1997; 2002) Deliberative Polls and Ned Cosby s Citizens Juries have been applied worldwide (Fung 2003; Hansen 2004). Deliberative democracy is heavily influenced by two schools of thought Rawlsian and Habermasian. The legacy and inspiration from these two theorists is apparent in many parts of the theory, but despite the work of Habermas and Rawls on deliberation, the concept of deliberation did not prosper until the late 1980s as figure 1 below indicates. Figure 1 measures the number of academic articles and book reviews published about deliberation in relation to democracy and gives an impression of the attention and popularity the concept of deliberative democracy has achieved. As figure 1 indicates, the academic debate on deliberative democracy has increased dramatically during the 1990s. From 1956 to 1995, only 14 academic articles were published in the field. From 1996 to 2005, 286 articles were published according to Social Science Citation Index. 1

3 Figure 1: Number of academic article on deliberative democracy Articles Book reviews Note: Articles and book reviews are found in "Social Science Citation Index" by searching for Deliberati* and Democracy in title, summary, and keywords in English language only. Publications from 1956 to 2005 are included in the figure. Search in SSCI conducted 4 April Why has the attention to deliberative democracy increased so dramatically by Looking at the figure, it certainly seems that it almost happened over night, but behind this shift one finds several explanations of why it happened. Firstly, one explanation is that within a few years, several articles and books independently of each other were published in relation to the subject of deliberation (e.g., Cohen 1989; Dryzek 1990; Fishkin 1991; Manin 1987; Miller 1992). Part of this explanation is also that the attention to Habermas work was revitalized by the translation in 1989 of his book from 1961: The structural transformation of the public sphere (Habermas 1989). Yet another explanation is due to a general trend among researchers to jump on the bandwagon. Thus, today deliberative democracy is a fashionable term, but that is not to say that the term is without content. On the contrary, deliberative democracy has been argued to bridge the liberal and republican tradition within democratic theory (Gutmann & Thompson 1996:27; Habermas 1996a). Deliberative democracy has also moved the rather entrenched debate between liberal and republican approaches to democracy, to a debate which seems to allow many approaches to democracy to find common grounds for their arguments. Despite the strong and increasing attention to deliberative democracy not much consideration has been given to the problems and contradiction in the theory. To bridge this gap in the research area this paper takes up the challenge to discus a paradox in the theory of deliberative democracy. The equality paradox in focus here argues that there is a strong tension within deliberative democracy as on the one hand unconstrained exchange of 2

4 argument is highlighted in the deliberative process and on the other hand the deliberative process gives priority to certain types of arguments. That is arguments that are based on reason and can be articulated on objective grounds. In this way the paper focuses on only a minor part of the complex theory of deliberative democracy. However, this narrow focus allows going into detail with one specific issue and, subsequently, elaborate on some empirical findings in relation to the paradox from a Danish national Deliberative Poll regarding whether Denmark should adopt the single European currency - the euro. The outline of the paper is twofold. First the equality paradox of deliberative democracy is discussed theoretically. Secondly empirical results from the Deliberative Poll highlight the paradox in a deliberative setting. The political equality paradox of deliberative democracy Political equality and liberty are often presented as the cornerstones of democratic theory. In reality, political equality and political liberty are most often understood as universal suffrage and freedom of expression, etc. In relation to deliberation, political equality and liberty are emphasized in the above definition of deliberation as deliberation involves an unconstrained exchange of arguments. Deliberation is, so to speak, a process where all participants during the deliberation should have an equal opportunity to express their opinions as they wish and are able to do so throughout the process. While this notion follows Dahl s ideal standard of Effective Participation (1989:109), it is often emphasized that deliberation will strengthen procedural legitimacy only as long as opinions are backed with reason (e.g., Habermas 1996b:448; Rättilä 2000). It is also argued that one of the advantages of deliberation is that it offers an advantageous position to arguments based on the common good and arguments with reference to general principles, compared to arguments based on self-interest and purely subjective arguments. The view that arguments stated in terms of the common good are advantageous in deliberation is also pinpointed by the notion of the force of the better argument, as some reasons are better than others in the sense that they are more convincing to all participants by referring to the common good and general interest. Accordingly, these types of arguments are more persuasive than other types (Habermas 1984; Manin 1987). Thus, deliberative democracy gives an a priori authority to certain kinds of reasons, which are defined prior to the deliberative process (Macedo 1999:3). Young (1996; 2000:37) has persistently criticized assumptions of some shared premises of speech culture or discursive 3

5 frameworks for being fallacies given the heterogeneity and plurality of society. Secondly, the effort to shape arguments according to some speech culture would exclude the expression of some views, needs, and injustices, because some participants may not be able to voice their claims according to such a framework. To be unable to communicate one s opinions according to a certain speech culture does not make these views and needs less relevant in the process, but accentuate that other forms of communication are needed in order to incorporate these views in the deliberative process. But is it possible to have a theory that on the one hand aims at political equality and on the other hand claims that some arguments are better than others? If all could equally express their interests through common good arguments and all could articulate the reasons for their opinions, it may be possible. However, critics of deliberation have argued, that it is far from the case as the asymmetries - in e.g. level of information and the participants capabilities in arguing - undermine any equality standard political equality requires. The deliberative process, critics of deliberative democracy argue, is not an equal process, as participants capable of arguing on rational, measurable and objective grounds are favored through the deliberative procedures. This challenge to the theory is also accentuated by the fact that such participants are already overrepresented in the political system (Phillips 1995; Sanders 1997). E.g. are elected representative systematically higher educated than the general public. Enhancing the deliberative aspect of a political system thus compromises political equality and favors reasoned, post-deliberative opinions and disfavors raw, emotional and top-of-the-head opinions. Besides being elitist in nature, this defining feature of deliberation also disadvantages people not used to express their opinions in terms of the common good. People inexperienced with meetings, less educated and shy people are easily sidetracked and, consequently, individuals with strong rhetorical skills and demagogues are in a more advantageous position when justifying their interests. The disadvantaged people s experiences and interests are lost in the process, thus creating a strong bias in the political system. The notion that there exist some universally acknowledged norms and conditions of how the deliberative processes must be conducted is basically Kantian and has received strong criticism from others. Bohman (1999) argues that the norms for the deliberative behavior develop in relation to the context, the social roles, and status which the participants possess. Thus, the procedures of deliberation should not be limited by an a priori set of universal norms, as these do not exist. Instead it is acknowledge the procedures of 4

6 deliberation develop in relation to and are embedded in the specific context and, thus, emerge according to the issues at stake. If deliberation was limited according to an a priori set of norms, the success of deliberation would be determined by whether the participants have the ability to abstract from their particular identities and social roles and not on their capacities to engage in deliberation with other participants. The a priori acknowledgment of certain norms for the deliberative procedure will, eventually, have the effect that the participants, who are unable to abstract from their social roles, are excluded from the deliberation, thus violating political equality not to mention liberty of speech and causing certain experiences to be lost in the deliberative process. This standpoint has been acknowledged by many deliberative democrats, who have emphasized that the norms for deliberation must develop within the group of deliberators during the process (Bohman 1996; Gutmann & Thompson 1996:224; Young 1996). Acknowledging that the procedures for deliberation develop between the participants during the deliberative process rather than being defined in advance according to some abstract principles, relaxes the tensions within the theory of deliberative democracy, but deliberative democracy still claims that the participants must be willing to enter the deliberative arena, in which they try to present their views, by appealing to reasons which the other participants can share. How exactly this is achieved, might be developed during deliberation, but it remains immanent in the theory that opinions, which can be presented in a mutually justifiable way, will receive a higher authority than other arguments referring to purely subjective interpretations of the situation. The political equality paradox as a self-contradiction in the deliberative democratic theory lie in that on the one hand most deliberative democrats strongly advocate deliberation for aiming at reasons referring to the common good, and on the other hand claim that deliberation is an unconstrained exchange of arguments. The tension exists because competing values simultaneously are emphasized in the theory. If true to the criterion of political equality it would challenge the aim of restricted common good deliberation - and vice versa. Being true to these competing and contradicting values in its ideal form thus becomes impossible. 5

7 Expanding the forms of deliberation Some have encountered the tension of deliberative democracy by calling on future designs of deliberative arenas to allow different types of communication to be a genuine part of the deliberative process and only excluding communication that completely lacks a respect for others or is incoherent (Young 2000:30). The three types of communication are greeting, rhetoric, and storytelling or narrative (Dryzek 2000; Young 1996; 2000). Greetings, verbal as well as nonverbal, such as gestures of flattery, deference and conciliatory caring should be included in the deliberation in order to stimulate trust among participants. Greetings, including small talk before the actual political discussion, are important because greetings are partly a public, explicit and mutual acknowledgment of each other. Without any form of greeting, deliberation becomes difficult as the participants have not recognized and shown the respect that is needed to establish the trust necessary for deliberation (Hedin 2001). Rhetoric is another form of communication which, according to Young (1996), must be allowed into the process. Allowing rhetoric would specify that an argument should appeal to the particular audience s experiences. It is also a way to get and keep the attention of the audience through the use of emotion, humor and wordplay, etc. which will increase the likelihood of reaching the desired outcome. Accordingly, the use of rhetoric opens up for claims that are embedded in the belief of the particular audience and acknowledges the audience s particular context, which not necessarily is framed in the common good but rather in the good of the audience. Narrative or storytelling is yet another form of communication, Young (1996; 2000) argues, is supplementary to other means of deliberation. Narrative involves a narrative description of the participant s particular experience and may thus help establish an intersubjective understanding of the situation. Narrative may also evoke sympathy and reveal the sources of the participant s values, which may serve to explain the underlying premises of a participant s opinion (Young 1996). Furthermore, narrative may help foster a shared understanding of the situation and allow experiences embedded in a particular belief or social role to be voiced. Such a shared understanding also provides the opportunity to share knowledge reflecting the particular characteristics of individuals, that otherwise might not have been voiced. Sharing such subjective experiences might enhance a shared understanding 6

8 of the situation, but it is also a way to encounter myths, biased interpretations and prejudices among the participants. However, it is necessary to point out that also these means of communication can be coercive or strongly manipulative, such as an emotional personal story or a rhetorical speech (Dryzek 2000), as Young also points out in her later work (Young 2000:78). Nevertheless, many deliberative democrats maintain that these forms of communication have to be genuine elements of the deliberative process in order to give the disadvantaged participants a better opportunity to articulate their experiences and to expand the possibility of a plurality of views surviving the deliberation (Dryzek 2000; Young 1996; 2000). Even though assigning more weight to these forms of communication in the deliberative process may help maintain a certain plurality, this claim is beside the main point. The point is that such means of communication are disadvantageous in the processes of exchanging arguments, not that these forms have not been allowed - deliberation is by definition an unconstrained exchange of arguments. The fundamental distinction, which frames deliberative democracy, is between communication based on reason and communication not based on reason, where the former is favored and the latter disadvantaged in the deliberative process. The three supplementary types of communication are also, as Young (2000) points out, inevitable parts of human interaction and must consequently also be present during deliberative processes. However, just by encouraging these means of communication does not make them more persuasive than before. Thus, deliberation still favors reasons with reference to the common good and arguments based on a broader understanding of the situation. Nevertheless, it would be fruitful to stress that these forms of communication are part of any deliberative process in order to facilitate a less biased process, but it remains embedded in the theory that these forms of communication are disfavored, compared to an argument based on a broad understanding of the issue. To include greeting, rhetoric, and narrative more directly in the deliberative process may contribute to expansion of the shared frame of interpretation and establishment of a deeper frame of reference, but these types of communication remain less convincing than an argument based on reason. The controversy about inclusiveness during deliberation The argument has so far rested on the implicit assumption that deliberation disadvantages some groups of people, who usually are underrepresented in the political system and, 7

9 accordingly, deliberation strengthens an already existing bias in the political system. Adapted from Young (2000) to general concepts of inclusiveness in relation to deliberation is analytic useful - external and internal inclusiveness. External inclusiveness refers to the ideal of representativeness. That is, do the participants mirror the population in general on opinions and socio-demographic characteristics? If certain citizens are excluded from the deliberative process and maybe even not allowed in the forum, external inclusiveness is violated. Internal inclusiveness refers to the ideal that when citizens debate an issue, all citizens should have an equal opportunity to express their views without having to commit to a certain framework or speech culture. A speech culture could be that arguments need not only to refer to some concept of the common good, but also that arguments need to be voiced in a rational and consistent way. If such a speech culture exists, impassioned and purely subjective understandings of the issue will have difficulties in the deliberative process. Furthermore, if citizens must commit to a certain speech culture when debating a political issue, where only certain kinds of arguments are allowed, it would disadvantage many citizens in the attempt to express their views. In such a case, deliberation violates internal inclusiveness. Furthermore, it is likely that any exclusion, due to the fact that a certain speech culture must be obeyed, reinforces already existing biases in a deliberative forum. That is, the groups of people which already are underrepresented in the deliberative forum due to a violation of external inclusiveness may also often be the groups which are affected by a violation of internal inclusiveness. The reasons for this reinforcement of a violation of internal inclusiveness during deliberation are twofold. First of all, the groups which are often underrepresented are also the groups of citizens which lack the capabilities to express their views in accordance with the speech culture. Secondly, if the speech culture is being maintained by the majority in the forum and if the forum is biased, the group may very well develop a speech culture which advances the majority, thus, excluding the already disadvantage groups. Due to its complexity, internal inclusiveness demands more attention. How can internal inclusiveness be measured? One way would simply be to find indicators of passiveness and domination during the deliberation. A more sophisticated measure would be not only to look at the quantity, but also at the quality of what is being said. Even though these indicators can be operationalized in an analytic context, they say nothing about the 8

10 passive participants= influence on the outcome of the deliberation. Silence and listening can be a quite effective tool of bargaining (Lewicki et al., 1994:195). If any biases in the internal inclusiveness are found, it raises yet another important question - why does this bias exist? One way to explore this relationship would be to analyze whether an unequal distribution of passiveness in the forum is distributed according to the degree that arguments referring to public interests dominate in the group. That is, it also has to be analyzed whether a strong speech culture affects the inclusiveness. Previous empirical claims on inclusiveness Research on external inclusiveness or representativeness is comprehensive and it is widely documented that some violation of external inclusiveness exists in most political forums. It is a fact that the population differs systematically from the elected representatives. Generally speaking, the elected representatives are e.g. better educated, older and more men than women are elected as representatives (Hansen et al. 2000; Narud & Valen 2000). In this way, it is well documented that the >usual suspects= or >gladiators= of political participation are biased in certain ways compared to the population in general. In particular the fact that women are underrepresented in political forums has received much attention (e.g., Phillips 1995). Wängnerud (2000) points out that in Sweden women MP=s prioritize gender equality and social welfare policies higher than their male colleagues. Phillips (1995) advocates the need for gender quotas in political forums in order to reduce the unequal distribution between the genders. Through gender quotas she hopes that the opinions of women will be adequately addressed in political forums. For several reasons, internal exclusiveness is not researched in the same degree. First of all it is much more problematic to measure internal inclusiveness. Measuring internal inclusiveness is problematic because finding reliable indicators in most cases involve much more interpretation and judgment. Secondly, it is harder to get access to data and, finally, it can be difficult to differentiate internal inclusiveness from external exclusiveness. E.g. is the reason why women speak less in parliament that they have a relatively minor share of important committee seats or is it because they are discriminated by an existing speech culture? In the existing research within the field, much research has focused on whether there is a relationship between the participants gender and their behavior during deliberation. 9

11 This has been the main critique raised against deliberative democracy inspired by feminist writings (Phillips 1995; Sanders 1997; Young 1996). However, only limited and weak empirical evidence has so far been presented. Research on focus group interviewing has suggested that internal inclusiveness is violated in some cases, even though no solid empirical data have been presented (Barbour & Kitzinger 1999; Morgan & Krueger 1993). Research on juries has also shown that men speak more and are more often selected as group leaders than women. Even when taking into account that men are more likely to be group leaders, men tend to dominate the discussion over women. Furthermore, studies of juries have shown than the more a person speaks the more likely it is that other group members regard this person as being persuasive and agenda setting of the discussion (Sanders 1997; Sunstein 2000). Also with regard to race, American classroom studies suggest that white people are more active and exert influence more than others during deliberation (Sanders 1997). In this way, the empirical findings indicate that the deliberative democratic ideal of an unconstrained exchange of argument is violated by actual deliberative processes because it seems that e.g. the jurors arguments are given unequal weight in the deliberative process. A study of private and public discussions indicate that women, the elderly and the poor are underrepresented among the highly active discussants and overrepresented among the passive discussants (Conover et al. 2002). Also Mansbridge (1999:363) cites several studies, which suggest that female US state legislators and women participants in Town Meetings speak relatively less and differently compared to their male colleagues. Sue Thomas (1994) on the other hand presents a different picture of the US=s state legislators as she finds no difference between men and women with regard to being active in the different legislative activities. On the other hand she confirms many other studies suggesting that women have different opinions with regard to gender issues. To sum up, even though the empirical findings of an exclusiveness potential of deliberation are mixed, the general picture suggests that deliberation in many cases can create some biases. In particular men seem to dominate the deliberation over women. Combining the findings indicates that the opinions of women are different from mens opinions, and the fact that women speak less suggest that a hierarchy within deliberative procedures exists, which in many cases favor certain views and disfavor other views. Therefore discrimination against minorities can be said to influence the exchange of arguments. If this is the case, deliberative 10

12 democratic theory is confronted with a challenge which questions the possibility of creating actual deliberative settings that fulfill the ideal of an unconstrained exchange of arguments. In the next section inclusiveness during the Danish Deliberative Poll is analyzed in detail. Inclusiveness during the Danish Deliberative Poll The national Danish Deliberative Poll conducted in August 2000 provides data to analyze deliberation in practice. The basic idea of Deliberative Polling is developed by Fishkin and his associates (Fishkin 1988; 1997; Hansen 2004). The idea is to bring together a representative group of people and let them deliberated among themselves and with leading politicians and experts. Throughout this process, their opinions were polled four times - before, at the beginning and at the end of the deliberative event, and finally three months later. During the weekend, they discussed the issue of the single European currency with each other in randomly composed groups of 18 to 20 lead by a trained neutral moderator, whose main task was to encourage and help everyone to take an active part in the deliberation and according to the participants evaluation, the moderators succeeded to perfection. In plenum, the participants asked questions to the leading political and experts. Before, during and after the weekend the participants were polled on attitudes, knowledge and assessment of the deliberative experience. For more details on the experiment and other analyses of the experiment see Hansen (2004). In relation to the representativeness the Poll succeeded to a large extent to bring together a representative group of people. Nevertheless, some biases where present e.g. women were somewhat underrepresented at the Poll, as is also the case in of representative of national parliaments. The bias is nevertheless quite small; if e.g. 30 participants had been women instead of men, the bias would have been reversed. However, the Poll in this case has somewhat replicated the biases found in many other political forums. In this context it must be mentioned that on occupation, place of living, and on sixteen out of seventeen opinion items underlying the euro opinion, there was no difference between the participants and the general population. Internal inclusiveness is related directly to the paradox for deliberative democracy. In order to provide indicators for the deliberative behavior within the 20 groups that deliberated during the weekend, the moderators assessed the participants of their groups 11

13 according to passiveness and other types of behavior. Another indicator of inclusiveness is the characteristics of the persons, who the groups decided were to raise their questions in the plenary sessions. In this way, inclusiveness is explored with two different indicators (see table 1). Table 1 shows that during the two days of deliberation, women are significantly more passive and act significantly less often as the expert type. 1 On the other hand there is no significant gender gap in those who are the most active in the groups or who raise the questions in the plenary sessions. Thus, the gender gap only exists among the most passive, but not among the most active participants. One interpretation of these findings is that the critique of deliberation from the feminists is justified. A deliberative process like this one shows a gender bias and, accordingly, this empirical, deliberative process does not live up to the deliberative ideal of an unconstrained process and thus certain experiences might be lost in the process. On the other hand, as there is no gender gap among the most active participants, it seems that the problem is less critical as women actively participate in the deliberation at the same rate as men. As the gender gap also exists with regard to the expert type suggests that part of the explanation is that there is an underlying self-confidence difference between men and women or that women are more modest during group deliberation. However, one thing might be the theoretical argument that the bias is a problem for the deliberative ideal, another is whether the passive women experience their passiveness as a problem. 1 The expert type was defined as a person who spoke up as being an expert, was convinced that what he or she said was correct and often seemed convincing to the other participants. 12

14 Table 1: The deliberative behavior of the participants in their groups divided on different characteristics (percentage, within group) Most Most Different type of behavior Raised questions in active passive Expert Dominant Shy Leadership plenary sessions Gender (n=364) - Men Women 27 27* 4* Education (n=357) - Lower education Upper education University degree 46** 10** 13* 10 8** 9 54** Age (n=364) years years years years 17* ** 7 29* Decisiveness (n=362) - Decided on the euro Non-decided on the euro 19* * Knowledge (n=363) - Low level of knowledge High level of knowledge 36** 12** 12** 9 11** 0** 51** Note: * indicates that the difference between the groups is significant at p<0.05 using a chi-square test for independence. ** indicate that the difference between the groups is significant at p<0.01 using a chi-square test for independence (2-sided). At the end of the weekend, the moderators were asked to point out the most active and most passive during the group deliberation and secondly asked to point out which of the participants could be characterized as an expert type, a dominant type, a shy type or as having shown leadership during the deliberation. A participant could easily be placed under more than one of these headings and sometimes not placed at all. The groups prepared a total of 226 questions for the panels in the four sessions. 161 different participants out of 364 were set to ask these questions. Decisiveness is measured at t1. The participants were asked how likely it was that they would change their vote before the referendum. Their answers were very unlikely, unlikely, neither nor and don t know. The "decisive" group includes the >very unlikely and >unlikely answers whereas other answers constitute the "non-decisive" group. Knowledge is measured at t1. The knowledge index combines 20 questions giving 5 points for each correct question. Thus, the index ranges from 0 to 100, where 100 indicate correct answers and 0 incorrect answers to the 20 questions (see Hansen, 2004: chapter 6 for details). An index value of 75 is the cut point between low and high knowledge giving approximately an equal number of participants in each group.

15 At the end of the weekend, the participants were asked in an open-ended question to tell about their experience in the deliberative process. In these questions none of the passive women described the process a being negatively, on the contrary strong enthusiasm was reported. Five women, rated as passive by their moderator, spontaneously reflected upon their passiveness during the deliberation. Their statements are included in table 2. The first four all emphasized their passiveness and immediately thereafter emphasized that they had learned much and gained knowledge by being part of the Deliberative Poll. From an individual level, the women benefited greatly from the experience. Table 2: Five passive women reflect on their concerns about being passive I have been very passive. I have listened and paid attention. I have learned much and gained much by listening. I am the type who does not say much in meetings. I used the time to listen. I gained much from being together with the groups, in spite of my silence. I have not argued for my opinions, but I have been a very keen listener, which has been very beneficial to me. The level of discussion was from the beginning very high. I did not know anything about the euro before the weekend and had difficulties participating in the discussion. Note: At the end of the weekend, the last question in the questionnaire asked the participants to reflect upon the fact that they had to express their opinions in front of other participants. The latter woman expressed a concern about being able actively to participate in the deliberation. The view supports the critique that deliberation is not an equal process, but clearly favors the more knowledgeable. To explore whether level of knowledge is the causing variable, table 1 also groups the participants with regard to their level of knowledge. It shows that level of knowledge seems to be an important factor with regard to the participants deliberative behavior. The more knowledgeable participants are more active, less passive, more often play the role as the expert or the leader and are less shy. Knowledge also plays a role when it comes to who raises the questions in the plenary sessions - 51% of the participants with a high level of knowledge raised a question whereas only 37% with a low level of knowledge did. In this way, knowledge seems to play a more important role than gender in the deliberative process. These findings contradict Conover et al. (2002), who finds a weak negative relationship between being an active political discussant and political knowledge. Conover et al. (2002:47) explain their findings by suggesting that perhaps their items to measure political 14

16 knowledge are irrelevant to the perception of level of political discussion or maybe the respondents with a low level of information actually engage in political discussions to gain information. The findings from the Danish Deliberative Poll can be interpreted as low factual knowledge being a barrier for active participation in the deliberative process. Nevertheless, the women=s spontaneous reaction to the deliberative experience also suggests that being a listener is not necessarily negative for the perception of the deliberative experience. The process of listening is an important part of the deliberative ideal in pursuit of mutually justifiable arguments, but if it is the case that the active participants also represent certain opinions, which are different from those of the passive participants, a bias may occur. Level of education, which also is included in table 1, shows significance on all groups except leadership and dominance. In particular when dividing the participants into the most active and most passive groups, the difference is quite large. The difference in percent between participants with a university degree and participants with lower education is 34 percentage points compared on being most active. Furthermore, the difference in percentage points between participants with lower education and participants with university degree compared on being most passive is 33. It is even more striking that a strong bias towards the higher educated also is present among the participants who raised the questions during the plenary sessions. Age also shows a significant relevance for understanding the groups behavior during deliberation, but the relationship is not linear as education. The age group between years is the most active and the young and the elderly participants are the most shy. The young participants passiveness during deliberation in the groups is in sharp contrast to the fact that the young participants most often raised the questions during the four plenary sessions. One explanation might be that it is less demanding for a participant to read out loud a question phrased collectively in the group than participating actively in the group discussion. To explore an alternative relationship, table 1 also includes whether the participants initially had decided what to vote in the upcoming referendum on the European single currency. One significant relationship is present. The more decisive participants were also the most active ones. Using a variant of an argument from social psychology, suggests that the more decided seek confirmation of their choice and, thus, find it easier to present a 15

17 consistent and one-sided argument in order to protect their choice compared to participants who are more in doubt which side to support. To sum up the findings: the most active participants in general have a university degree, are between years old, are decisive with regard to their voting intentions and have a high level of factual political knowledge on the euro issue, whereas the most passive participants are the females, the lower educated and participants with low factual knowledge on the issue. The participants with a lower education, the young, the elderly and the participants with low factual knowledge are more often classified as being the shy types. However, the young participants are the most active with regard to asking questions to the politicians and experts in the plenary sessions. A clear bias in the deliberative processes is reported and the bias is even present during the setting where the moderator helped and encouraged equal participation. It may, therefore, seem plausible that in other deliberative settings, where a moderator is not present, this bias will be emphasized. Secondly, as Denmark is a relatively highly homogenous society, it is even more significant that the bias in the deliberation activities is relatively strong. In less homogenous societies, these differences might very well be even more marked. Even though the analysis clearly shows that an internal inclusiveness bias exists during the Danish Deliberative Poll, the results are not as uninspiring from a deliberative democratic point of view as it might look. Many of the variables: education, decisiveness, and factual knowledge are not simply a matter of inherited abilities, but rather abilities which people can achieve throughout life. Thus, by raising the level of education it is possible to reduce the bias in the inclusiveness. Was the bias entirely a matter of gender, the situation would have been different. The analysis presented in table 1 does nevertheless only present a bivariate analysis and, in this way, does not allow us to conclude on the relative strengths of the independent variables. In order to draw such conclusions, a multivariate analysis is needed. Such an analysis is presented in table 3. 16

18 Table 3: The effect of different variables on the participants= deliberative behavior (binomial logistic regression, odd ratios) Different type of behavior Most active Most passive Expert Dominant Shy Leadership Raised questions in plenary sessions Constant 0.029** 3.670* 0.003** 0.003** 3.998* 0.009** Gender (1) * 0.361* Education (2) 3.018** 0.404** 2.120* 2.171* 0.405** ** Age (3) 0.629** ** Decisiveness on the euro (4) High level of knowledge (5) Hosmer and Lemeshow=s goodness-of-fit test 2.446** ** 0.391** 2.933* ** 3.881*.468 X 2 = Sig.=0.151 X 2 = X 2 = Sig.=0.283 Sig.=0.698 X 2 = Sig. = X 2 = Sig. = X 2 = Sig. = X 2 =9.942 Sig. = Nagelkerke R Square Correctly predicted 74% 81% 92% 94% 81% 94% 58% n Note: * significant at 0.05 level. ** significant at 0.01 level. (1) Women compared to men. (2) Level of education is measured as 1 = lower education, 2 = upper education, and 3 = university degree. (3) Age is measured by four groups 1 = years, 2 = years, 3 = years, and 4 = 61+ years. (4) Decisive compared to non-decisive. (5) Participants with a high level of knowledge compared to participants with a low level of knowledge. See table 13.2 for details on the variables. The bold odd ratio in each regression has the largest significant impact on the dependent variable. Generally speaking, table 3 confirms the bivariate analyses from table 1. The discussion will focus on the first two regressions in table 3. Table 3 shows that the odds ratio of being active triples by a unit increase in education. Furthermore, being decisive or having a high level of knowledge more than doubles the likelihood of being active in the deliberation. Age is also found to have a significant effect. The younger participants are generally more speaking and are likely to be active. However, it is worth to notice the result from table 3 indicating that age did not have a linear effect on the participants deliberative behavior as the participants who were years old were the most active. Being a woman increases the likelihood of being passive in the deliberation by a factor of 1.8, whereas an increase in the level of education increases the likelihood of being passive by a factor of 0.4, that is, an increase in the level of education reduces the likelihood of being passive by a factor of 2.5 (the reciprocal value of 0.404). A high level of knowledge reduces the chance of being passive during deliberation by a factor of 2.6 (the reciprocal value of 0.391). In this way, when controlling for age, education, decisiveness and level of 17

19 knowledge, gender still comes out as significant, but both education and knowledge have a larger impact on whether a participant is passive during deliberation. In none of the seven regressions in table 3 does the gender issue have the largest impact on the participants deliberative behavior. On the other hand, knowledge and education are found to have a more consistent and important impact on participants= behavior. A general conclusion must be that internal inclusiveness is violated during the deliberative process, as women, the less educated, participants with a low level of factual knowledge are more passive than anybody else. In this way, an empirical indication of a violation of political equality and internal inclusiveness in a deliberative setting is provided. Nevertheless, education and knowledge turn out to be the most important factors for the participants deliberative behavior and not gender which although often is the highlighted variable. A first step to understand the exclusiveness potential of deliberation is analyzed. But yet another question needs to be explored. The general concern of the feminist writers was that the bias in deliberation tends to exclude certain views in the process. To understand the relationship, table 4 divides the participants opinions at the end of the Deliberative Poll with regard to level of activeness and passiveness during the deliberation. The results presented in table 4 are quite interesting. On all questions there are significant opinion differences between the most active and the other participants. The most active participants are much stronger advocates of the single currency than other participants, whereas the most passive are much more skeptical toward introducing the single currency in Denmark. Accordingly, the data from the Deliberative Poll on the euro show a strong bias in the deliberation toward a pro-european view. A first interpretation is that the strong bias in the political campaigns up to the referendum toward pro-euro views is replicated in the deliberation during the Deliberative Poll. The referendum campaign was dominated by a massive yes-campaign, which was mainly run on supposedly objective and rational arguments focused on economic issues and more influence in EU-matters. The pro-euro participants may, accordingly, find it easy just to echo this discourse and follow this line of the massive yes-campaign. On the other hand, many of the advocates supporting a no to the euro partly rested their arguments on fear of loss of national identity and cultural tradition (Buch & Hansen 2002). 18

20 Table 4: Mean difference on eight opinion items and the voting intention divided on deliberative behaviors The most active participants compared to other participants The most passive participants compared to other participants It is an important priority for the EU to include additional countries in the Union as soon as possible 13.50** Danish participation in the single currency reduces Denmark s independence ** 4.55 Danish participation in the single currency lowers the current Danish interest rates 12.22** ** Danish participation in the single currency weakens the Danish welfare system ** 4.55 The single currency is a step toward >The United States of Europe= ** 5.65 Danish participation in the single currency would give Denmark a stronger say in EU decisions 10.32** Danish participation in the single currency would pose a threat to the Danish national feeling ** 15.53** Danish participation in the single currency would create better conditions for the Danish business 6.89** community Voting intentions with regard to the euro 18%** -7% Note: n for the active group is whereas other participants have an N of n for the passive group is 71-76, whereas other participants have an n of The mean is calculated on a scale where strongly agree = 100, somewhat agree = 75, neither agree, nor disagree and don t know = 50, somewhat disagree = 25 and strongly disagree = 0. Thus, a positive difference indicates that the most active/passive participants agree more with the statement than other participants. The percentages of the voting intention indicate how many percentage points pro-euro participants have over the euro-skeptical participants within each group (active/passive) compared to other participants. The opinions are polled at the end of the weekend, using the poll before the weekend does not change the results significantly or consistently. ** indicate that the mean difference is significant at p<0.05. (2-tailed test).

21 The two different discourses among the participants follow the reasoning in deliberative democracy that arguments that are backed with reason and referring to general interests and the common good - such as improvement of the national economy - should be favored. The argument resting on national identity is much more difficult to present in a persuasive way so that all deliberators find it a good argument. Improved national economy is on the other hand more likely to find general support. Thus, the bias in table 4 supports the tension of political equality, that arguments referring to objective rational reasons seem to be favored in deliberation at the expense of a feeling e.g. toward being Danish. Another part of the explanation of this bias in activism is a slightly overrepresentation of more pro-european participants compared to skeptical participants, but this difference cannot account for the large differences. Furthermore, the relatively large bias in the activism did not result in a massive change of opinion toward more pro-european views during the deliberation as the number of participants indicating to vote yes to the euro gained relative 24 percent, whereas the relative gain in favor of a no was 16 percent (see Hansen, 2004:chapter 7). There was no demand on the groups to reach a collective decision at the end of the weekend which is probably why the disproportionate activism did not result in opinion changes toward more pro-european opinions. However, it is striking that although the groups did not have to reach a collective decision and as the moderators helped and encouraged everyone to participate in the process, the bias in activism is still reported. Accordingly, in a deliberative setting with a demand for collective decisions, and not moderated, the violation of internal inclusiveness may very well be even more prevalent. It has been argued that violation of internal inclusiveness partly is due to the fact that a certain speech culture evolves during deliberation and that such a speech culture favors certain participants. So far, it has been shown that the participants socio-demographic characteristics are related to exclusion during deliberation, but is the exclusion also related to the degree to which the groups commit themselves to argue with reference to general principles and the common good? An indicator of the prevalence of a certain speech culture in the groups during the Deliberative Poll is the participants answers to two questions regarding the deliberative process. Both questions ask whether the participants in their group argued with reference to general principles and what is thought best for all citizens. Generally speaking, the 20